Identification of pancreatic cancer stem cells and selective toxicity of chemotherapeutic agents.

Background & Aims: Identification and purification of cancer stem cells (CSCs) could lead to new therapeutic targets, but their heterogeneous expansion is an obstacle to their study. We investigated whether it is possible to monitor pancreatic CSCs in real time, based on their intrinsic low level of proteasome activity.

Methods: We engineered human pancreatic adenocarcinoma cells (PANC1, MIAPaCa2, BxPC3, and KLM1) to express a green fluorescent molecule fused to the degron of ornithine decarboxylase (Gdeg) from a retroviral vector; the fluorescent Gdeg accumulates in CSCs as a result of low activity of the 26S proteasome. Cells with high and low levels of fluorescence (Gdeg(high) and Gdeg(low)) were isolated by flow cytometry; tumor growth was analyzed in immunocompromised mice. We performed a screen for agents that were specifically toxic to pancreatic CSCs, in a synthetic lethal manner.

Results: Gdeg(high) cells, but not Gdeg(low) cells, formed spheres and underwent asymmetric division-features of CSCs. Injection of as few as 10 Gdeg(high) cells led to tumor formation in mice. Gemcitabine was toxic to cultured Gdeg(low) cells, whereas Gdeg(high) cells were resistant. We observed that quercetin was toxic to Gdeg(high) cells in culture and in pre-established tumors grown from these cells in mice. Nuclear accumulation of β-catenin was detected in Gdeg(high), but not Gdeg(low), and lost after exposure to quercetin.

Conclusions: We used a fluorescence marker system for level of proteasome activity to identify pancreatic cancer cells with features of cancer stem cells. We identified quercetin as a compound that is specifically toxic to pancreatic CSCs.